Every article reads without JavaScript. Search and the newsletter form need it; plain-text editions are at /index.md and /llms.txt.

Models

25

Pieces

OCT 28, 2026

Last filed

pricing open-weights benchmarks openai frontier-models china

Gemini 4 Carbon takes the top of the leaderboard as a model only vetted cyber defenders can use

Google's Gemini 4 Carbon took first place on Artificial Analysis's index on October 28 with a score of 59, but it launched only to vetted cyber defenders through the Fairwind Program, repeating the staged release Google used for Argon four weeks earlier.

Models · Oct 28 · 4 min read · 6 sources

Read the piece →

Also in Models

October 17, 2026

Frontier on a desktop: the weights fit, but the price cap comes from the cloud

Seven weeks of open-weight releases put models within a benchmark generation of the frontier on a $5,499 Mac Studio or a 12 GB gaming GPU, but the measured numbers show batched cloud hosts, not desktops, now set the price of mid-market tokens.

October 16, 2026

Jev named a new model category, and its interface was cloned in four weeks

TypeSafe's Jev made fast, calibrated decision models the first new model category since reasoning models, but within four weeks its request format ran on Ollama, Cloudflare, Microsoft and 182 open-weight systems, while Jev earned 0.27% of Vercel's gateway spend.

Everything in Models

25 pieces · newest first

Three labs, six weeks, one tie: grading the autumn frontier against independent reruns

Anthropic, OpenAI and Google shipped new flagships between September 1 and October 13, and once independent evaluators reran the headline benchmarks the frontier compressed into a statistical tie, leaving cost, tokens, speed and access as the differences buyers can measure.

GPT-6.1 Sol costs a fifth of Astra. Fast Sol costs more than Astra.

GPT-6.1 Sol ties GPT-6 Astra on OpenAI's own coding benchmark at $2 and $10 per million tokens, a fifth of Astra's rate, but its Ultrafast tier costs $12 and $60, which moves OpenAI's price premium from intelligence to latency.

Opus 5.5 and GPT-6 Sol shipped the same day, and neither launch table could see the other

Anthropic released Claude Opus 5.5 at $4 and $20 per million tokens and OpenAI released GPT-6 Sol at $2 and $10 about 90 minutes apart on September 22, and because each launch table benchmarks the other lab's previous model, the price sheets are the only like-for-like comparison.

TypeSafe's Jev gives up text to sell probabilities at $0.042 per million tokens

TypeSafe's Jev answers typed questions with probabilities instead of generated text, which lets it charge $0.042 per million input tokens with free output, but its headline speed and cost multiples come from four workflows the company built itself.

GPT-6 Astra costs 2.5x GPT-5.6 Sol, and its best cyber scores belong to a model users can't run

OpenAI released GPT-6 Astra on September 3 at $10 and $50 per million tokens, 2.5 times GPT-5.6 Sol's promotional rate, as the first model it rates Critical in cybersecurity, and several of its biggest launch numbers describe configurations buyers do not get.

Claude Fable 5.1 cuts the cache price 75%. At max effort, a task still costs more.

Anthropic shipped Claude Fable 5.1 to everyone and the same weights as Mythos 5.1 to vetted US organizations, cutting cache reads 75% to $0.25, but the first independent run found a max-effort task costs 20% more than on Fable 5.

GLM-5.3 goes open weight, and Z.ai's best model leaves MIT behind

Z.ai published the 753-billion-parameter GLM-5.3 on Hugging Face two weeks after its API launch, but swapped GLM-5.2's MIT license for one that makes model hosts above $10 billion in revenue pass a Z.ai security review.

Post-training is the new pre-training

Grok 4.6, Gemini 3.7 Flash, DeepSeek V4-Pro-0813, and GLM-5.3 shipped within three days of each other and not one is a new base model; the frontier is advancing by post-training, harness, and serving efficiency while every next-generation base stays in the oven.

The toll booth on the open road: grading August's open licenses

Qwen3.8-Max's weights went live today under a license that charges companies above $50M in revenue, Kimi K3's requires a separate deal above $20M (reportedly at up to 30%), and Meta's Glimmer is Apache-clean: 'open weight' just fractured into regimes of who pays.

Zuckerberg's 6,500-word bet: Meta re-enters the giveaway

Meta returned to open weights with Muse Glimmer, a 30B model under Apache 2.0 wrapped in a manifesto arguing that distribution is safety, five days after disclosing its own model had hacked another company, and the essay is best read against that timeline.

The model that cut its own price

OpenAI cut GPT-5.6 Luna by 80% three weeks after launch, saying Sol rewriting its own production kernels funded the cut, which makes the real story less a blink at China than a model becoming a line item on both sides of its own P&L.

Half the price of frontier

Anthropic shipped Claude Opus 5 as near-frontier intelligence at half the cost of Fable 5, and read from the P&L, the launch is a margin admission that the race has moved from raw capability to cost-per-task.

Google shipped three Flash models the day its flagship went missing

Google's July 21 drop cut Flash token prices 17% and added a security-tuned Cyber variant, but the tell is the model that didn't ship: Gemini 3.5 Pro, promised for June, is still partner-only while compute moves to Gemini 4.

One fortnight, four open frontiers: grading China's open-weight surge

In roughly fourteen days, four Chinese labs shipped or previewed near-frontier open-weight models (LongCat-2.0, Kimi K3, Qwen 3.8-Max, GLM-5.2), and because 'open' is a spectrum being gamed, this map grades actual openness, not press-release openness.

Kimi K3 and the three-trillion-parameter open ceiling

Moonshot AI shipped a 2.8-trillion-parameter model it calls the world's first open 3T-class model while reportedly raising at a $30 billion valuation, and the interesting part is not the parameter count: a startup pitching scarcity promised to give its best asset away.

Three labs shipped. One asked permission.

In 72 hours in early July, SpaceXAI shipped Grok 4.5, Meta shipped Muse Image and Muse Spark 1.1 and OpenAI shipped GPT-Live while GPT-5.6 Sol waited on a government review, and the real frontier, agentic coding at $2 a million tokens, routed straight around that gate.

The gate and the giveaway

While Washington spent June turning frontier model releases into licensed events, Meituan open-sourced LongCat-2.0, a 1.6-trillion-parameter agentic-coding model under an MIT license trained entirely on domestic chips, as the counter-move to the export-control regime.

Google just priced the generative-media floor: $0.034 an image

Google shipped Nano Banana 2 Lite at $0.034 per image and Gemini Omni Flash at $0.10 per second of video on the same day, setting a public floor price for high-volume generative media low enough to change who can afford to run it.

The open-weight reasoning gap is now 3.4 months, and the math is public

On June 16, GLM-5.2 became the leading open-weight model on the Artificial Analysis Intelligence Index v4.1 at 51, just 3.4 months behind an equally capable closed model, down from a 9.8-month gap in December 2024.

The open-weight coding frontier caught Claude, and it speaks Mandarin

Within three weeks MiniMax M3, Qwen3.7-Max, DeepSeek V4, Kimi K2.6 and Nemotron 3 Ultra all claimed the coding frontier, and third-party leaderboards show a real but small gap, mostly self-graded marquee numbers and a jurisdiction risk no benchmark measures.

Google I/O '26 decoded: the agentic era, 3.2 quadrillion tokens, and the Pro that didn't ship

Google's 2h57m I/O 2026 keynote landed eight first-class platform shifts, from Gemini 3.5 Flash and Antigravity 2.0 to TPU 8t/8i and a $180-190B 2026 capex line, while the Pro model didn't ship.

The voice-agent stack just collapsed

Three production speech-to-speech APIs shipped in a single quarter, and with them the cascaded STT→LLM→TTS pipeline that voice agents were built on has become a legacy architecture.